AI-Based DL Model Shows Promise in Breast Cancer Detection
Artificial intelligence (AI) and deep learning (DL) advancements have ushered in new possibilities for improving medical diagnostics and patient care. A recent study published in Radiology: Artificial Intelligence has highlighted the potential of an AI-based DL model in detecting precancerous changes in women at high risk for breast cancer. This research holds significant promise for enhancing breast cancer detection and risk stratification, particularly in populations with a heightened susceptibility to the disease.
The study focused on utilizing a DL model that was trained on a vast dataset of screening mammograms. By measuring its predictive accuracy using the area under the receiver operating characteristic curve (AUC), the performance of the DL model was evaluated. The results have shown promising outcomes, with the DL model achieving a 71 percent AUC for one-year breast cancer prediction and a 65 percent AUC for a five-year prediction. Comparatively, the traditional Breast Imaging Reporting and Data System (BI-RADS) system had a slightly higher one-year AUC at 73 percent, but the DL model outperformed it for long-term breast cancer prediction with a five-year AUC of 63 percent, surpassing BI-RADS’ 54 percent.
The study also investigated the role of imaging in predicting future cancer development by conducting mirroring experiments. These experiments aimed to assess the accuracy of the DL model in detecting early or premalignant changes that may not be visible in standard mammograms. The results indicated the significance of imaging the breast with future cancer in influencing the DL model’s performance. Positive mirroring yielded a 62 percent AUC, while negative mirroring showed a 51 percent AUC, highlighting the DL model’s potential in detecting premalignant or early malignant changes.
One particularly promising finding was the potential for the DL model to complement the BI-RADS system in short-term risk stratification. By combining the results of the DL model with BI-RADS scores, discrimination was enhanced, suggesting that DL tools could improve the assessment of screening mammograms and provide more accurate predictions for near-term risk assessment.
To maintain scientific integrity, the researchers noted that the DL model’s training dataset consisted of high-risk women with lower-risk profiles. Therefore, caution should be exercised in directly applying these findings to women at average risk for breast cancer. Further research is needed to explore the DL model’s applicability in diverse populations and its potential to aid in breast cancer detection and risk assessment for a broader range of patients.
Overall, this study emphasizes the significant promise of DL models in breast cancer detection and risk stratification, particularly for individuals at high risk. It sets the stage for future research to refine DL models, expand their utility to diverse populations, and ultimately contribute to improved breast cancer diagnosis and patient outcomes. As technology advances, AI-driven solutions have the potential to revolutionize breast cancer screening and management, leading to earlier detection and enhanced patient care.
